The epidemiology of acute paediatric osteoarticular infections (OAI) has recently evolved, mainly due to the improvement of microbiological diagnosis. We conducted a prospective study to analyse the recent epidemiology and the clinical evolution of paediatric OAI in order to validate the adequacy of our probabilistic first-line antibiotic treatment (intraveinous cefamandole + gentamicin). All children suspected of community acquired OAI were included and followed-up for 3 years. The etiologic diagnosis was based on blood cultures, joint aspirations and bone punctures. All osteoarticular (OA) samples were systematically inoculated into blood culture bottles. Real-time universal 16S rRNA and PCR targeted on Staphylococcus aureus, Kingella kingae, Streptococcus pneumoniae and Streptococcus pyogenes were performed twice a week. From 17 March 2007 to 26 February 2009, 98 septic arthritis, 70 osteomyelitis, 23 osteoarthritis and six spondylodiscitis were analysed. A portal of entry was suspected in 44% of cases, including 55% of otorhinolaryngological infections. C reactive protein was the most sensitive inflammatory marker. PCR increased by 54% the performance of bacteriological diagnosis. Among the patients completely investigated (blood culture and OAI samples), there were 63% documented OAI. The main pathogens found were K. kingae (52%), S. aureus (28%), S. pyogenes (7%), S. pneumoniae (3%) and Streptococcus agalactiae (2%). All isolated bacteria were sensitive to the probabilist treatment and outcome was favorable. PCR has significantly improved the performance and the delay of IOA diagnosis in children, for which K. kingae turned out to be the first causative agent. The probabilistic treatment was active against the main bacteria responsible for paediatric OAI.